INDEX

Explanations

the word "Sab" with varying levels of accuracy

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Sab" or variations of it

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ä»£

-0.76

ä¸ī

-0.73

ä

-0.73

äºº

-0.72

uminati

-0.72

opathy

-0.71

aturally

-0.67

ä¹ĭ

-0.67

opathic

-0.67

culosis

-0.66

POSITIVE LOGITS

Sab

0.94

Sab

0.90

rina

0.90

aram

0.84

ras

0.79

rill

0.77

etr

0.77

onis

0.77

dar

0.76

ril

0.76

Activations Density 0.003%

the word "Sab" with varying levels of accuracy

mentions of the name "Sab" or variations of it

No Comments

No Known Activations